Copy-move Forgery Detection Based on LPP and Improved SIFT Algorithm

Image copy-move forgery detection is an important part of image forgery detection. In this paper, we propose a novel forgery detection algorithm based on the SIFT and LPP algorithm. First,we use LPP algorithm to reduce the dimension of feature points and feature-vectors which are generated by the SIFT algorithm.Solving many defects of the traditional SIFT algorithm like the number of feature points are too many,the dimension of the feature-vector is too high etc. Then we cluster the similar feature points using the cohesive hierarchical clustering algorithm to find out the copy-move forgery area. At the end of the article, we make some experiments to test our algorithm,using 100 pictures in the Columbia University copy-move forgery image library. Results show that the proposed algorithm can generate fewer feature points and lower dimensions of the feature-vector than traditional SIFT, SURF, and PCA-SIFT algorithm,making the efficiency and the accuracy rate of the image forgery detection greatly improved,regardless of how the forgery regions are stretched or rotated.